* lib/rss/parser.rb: added entity handling type predicate.

* lib/rss/rexmlparser.rb: ditto.
* lib/rss/xmlparser.rb: ditto.
* lib/rss/xmlscanner.rb: ditto.

* lib/rss/xmlscanner.rb: more robust entity handling.

* test/rss/test_parser.rb: added an entity handling test.


git-svn-id: svn+ssh://ci.ruby-lang.org/ruby/trunk@9602 b2dd03c8-39d4-4d8f-98ff-823fe69b080e

ChangeLog

@@ -1,3 +1,14 @@
+Wed Nov 23 22:34:15 2005  Kouhei Sutou  <kou@cozmixng.org>
+
+	* lib/rss/parser.rb: added entity handling type predicate.
+	* lib/rss/rexmlparser.rb: ditto.
+	* lib/rss/xmlparser.rb: ditto.
+	* lib/rss/xmlscanner.rb: ditto.
+
+	* lib/rss/xmlscanner.rb: more robust entity handling.
+
+	* test/rss/test_parser.rb: added an entity handling test.
+
 Wed Nov 23 20:59:01 2005  Hidetoshi NAGAI  <nagai@ai.kyutech.ac.jp>
 
 	* ext/tk/lib/tk.rb: add Tk.pkgconfig_list and Tk.pkgconfig_get

lib/rss/parser.rb

@@ -69,7 +69,6 @@ module RSS
         parser.ignore_unknown_element = ignore_unknown_element
         parser.parse
       end
-
     end
 
     def_delegators(:@parser, :parse, :rss,
@@ -113,8 +112,14 @@ module RSS
 
   class BaseParser
 
+    class << self
+      def raise_for_undefined_entity?
+        listener.raise_for_undefined_entity?
+      end
+    end
+    
     def initialize(rss)
-      @listener = listener.new
+      @listener = self.class.listener.new
       @rss = rss
     end
 
@@ -205,6 +210,10 @@ module RSS
         def_get_text_element(uri, name, *get_file_and_line_from_caller(1))
       end
       
+      def raise_for_undefined_entity?
+        true
+      end
+    
       private
 
       def def_get_text_element(uri, name, file, line)

lib/rss/rexmlparser.rb

@@ -10,12 +10,13 @@ module RSS
   
   class REXMLParser < BaseParser
 
-    private
-
-    def listener
-      REXMLListener
+    class << self
+      def listener
+        REXMLListener
+      end
     end
-
+    
+    private
     def _parse
       begin
         REXML::Document.parse_stream(@rss, @listener)
@@ -35,6 +36,12 @@ module RSS
     include REXML::StreamListener
     include ListenerMixin
 
+    class << self
+      def raise_for_undefined_entity?
+        false
+      end
+    end
+    
     def xmldecl(version, encoding, standalone)
       super(version, encoding, standalone == "yes")
       # Encoding is converted to UTF-8 when REXML parse XML.

lib/rss/xmlparser.rb

@@ -59,11 +59,13 @@ module RSS
 
   class XMLParserParser < BaseParser
 
-    private
-    def listener
-      XMLParserListener
+    class << self
+      def listener
+        XMLParserListener
+      end
     end
-
+    
+    private
     def _parse
       begin
         parser = REXMLLikeXMLParser.new

lib/rss/xmlscanner.rb

@@ -1,19 +1,29 @@
 require 'xmlscan/scanner'
+require 'stringio'
 
 module RSS
   
   class XMLScanParser < BaseParser
     
-    private
-    def listener
-      XMLScanListener
+    class << self
+      def listener
+        XMLScanListener
+      end
     end
-
+    
+    private
     def _parse
       begin
-        XMLScan::XMLScanner.new(@listener).parse(@rss)
+        if @rss.is_a?(String)
+          input = StringIO.new(@rss)
+        else
+          input = @rss
+        end
+        scanner = XMLScan::XMLScanner.new(@listener)
+        scanner.parse(input)
       rescue XMLScan::Error => e
-        raise NotWellFormedError.new(e.lineno){e.message}
+        lineno = e.lineno || scanner.lineno || input.lineno
+        raise NotWellFormedError.new(lineno){e.message}
       end
     end
     
@@ -57,7 +67,7 @@ module RSS
     end
 
     def on_entityref(ref)
-      text(ENTITIES[ref])
+      text(entity(ref))
     end
 
     def on_charref(code)
@@ -79,7 +89,7 @@ module RSS
     end
 
     def on_attr_entityref(ref)
-      @current_attr << ENTITIES[ref]
+      @current_attr << entity(ref)
     end
 
     def on_attr_charref(code)
@@ -97,6 +107,15 @@ module RSS
       tag_end(name)
     end
 
+    private
+    def entity(ref)
+      ent = ENTITIES[ref]
+      if ent
+        ent
+      else
+        wellformed_error("undefined entity: #{ref}")
+      end
+    end
   end
 
 end

test/rss/test_parser.rb

@@ -122,7 +122,20 @@ EOR
         end
         assert_parse(rss, :nothing_raised)
       end
+    end
 
+    def test_undefined_entity
+      return unless RSS::Parser.default_parser.raise_for_undefined_entity?
+      assert_parse(make_RDF(<<-EOR), :raises, RSS::NotWellFormedError)
+#{make_channel}
+#{make_image}
+<item rdf:about="#{RDF_ABOUT}">
+  <title>#{TITLE_VALUE} &UNKNOWN_ENTITY;</title>
+  <link>#{LINK_VALUE}</link>
+  <description>#{DESCRIPTION_VALUE}</description>
+</item>
+#{make_textinput}
+EOR
     end
 
     def test_channel